R Notebook

library(spatialLIBD)
library(scuttle)
library(dplyr)
library(ggcorrplot)
library(ggplot2)
library(EBImage)
library(ggpubr)

library(foreach)
library(doParallel)
registerDoParallel(8)

Get Histology Image and Plot

# get histology image
img <- SpatialExperiment::imgRaster(spe_sub)

## Transform to a rasterGrob object
grob <- grid::rasterGrob(img, width = grid::unit(1, "npc"), height = grid::unit(1, "npc"))

## Make a plot using geom_spatial
p <- ggplot2::ggplot(
    sample_df,
    ggplot2::aes(
        x = pxl_col_in_fullres * SpatialExperiment::scaleFactors(spe_sub),
        y = pxl_row_in_fullres * SpatialExperiment::scaleFactors(spe_sub),
    )
) +
    geom_spatial(
        data = tibble::tibble(grob = list(grob)),
        ggplot2::aes(grob = grob),
        x = 0.5,
        y = 0.5
    )

## Show the plot
print(p)

png_link <- imgData(spe)[imgData(spe)$sample_id == sample_id,]$data[1]
img <- readImage('https://spatial-dlpfc.s3.us-east-2.amazonaws.com/images/151507_tissue_lowres_image.png')
#display(img)

red <- imageData(EBImage::channel(img, 'red'))
green <- imageData(EBImage::channel(img, 'green'))
blue <- imageData(EBImage::channel(img, 'blue'))
grey <- imageData(EBImage::channel(img, 'grey'))
hist(red)

hist(green)

hist(blue)

hist(grey)

display(red)

display(green)

display(blue)

# create boolean array for logcounts
logcounts_151507 <- logcounts(spe_sub)

# transpose logcounts matrix
logcounts_151507 <- t(logcounts_151507)

# remove columns with 0 max gene expression vals across all barcodes
zv <- apply(logcounts_151507, 2, function(x) length(unique(x)) == 1)
logcounts_151507_filtered <- as.matrix(logcounts_151507[,!zv])

dim(logcounts_151507)

[1]  4165 33538

dim(logcounts_151507_filtered)

[1]  4165 21134

Create DF Holding Pixel locations and Pixel Color Channel Values

# get scale factor
scaling <- SpatialExperiment::scaleFactors(spe_sub)

# get names of barcodes in sample, filter pxl dataframe to only include those barcodes
spatial_coords <- spatialCoords(spe_sub)

# create data frame
data <- data.frame(spatial_coords, row.names=NULL)
data['barcode'] <- rownames(spatial_coords)

# add scaled pixel columns
new <- ceiling(data[["pxl_col_in_fullres"]] * scaling)
data[ , ncol(data) + 1] <- new 
colnames(data)[ncol(data)] <- "scaled_pxl_col_in_lowres"

# add scaled pixel rows
new <- ceiling(data[["pxl_row_in_fullres"]] * scaling)
data[ , ncol(data) + 1] <- new 
colnames(data)[ncol(data)] <- "scaled_pxl_row_in_lowres"

# create vectors to be added to dataframe
X_r <- vector(mode="double", length = dim(data)[1])
X_g <- vector(mode="double", length = dim(data)[1])
X_b <- vector(mode="double", length = dim(data)[1])
X_grey <- vector(mode="double", length = dim(data)[1])

# iterate over dataframe, add rgb values
for (i in 1:nrow(data)) {
    col_index <- data[i,]$scaled_pxl_col_in_lowres
    row_index <- data[i,]$scaled_pxl_row_in_lowres
    X_r[i] <- red[col_index, row_index]
    X_g[i] <- green[col_index, row_index]
    X_b[i] <- blue[col_index, row_index]
    X_grey[i] <- grey[col_index, row_index]
}

data[ , ncol(data) + 1] <- X_r
colnames(data)[ncol(data)] <- "red"
data[ , ncol(data) + 1] <- X_g
colnames(data)[ncol(data)] <- "green"
data[ , ncol(data) + 1] <- X_b
colnames(data)[ncol(data)] <- "blue"
data[ , ncol(data) + 1] <- X_grey
colnames(data)[ncol(data)] <- "grey"

dim(data)

[1] 4165    9

data

Create ind_barcodes DF only if each spot has multiple pixels associated so that mean/median/modecan be computed (For HighRes Images)

# Mode Function
getmode <- function(v) {
   uniqv <- unique(v)
   uniqv[which.max(tabulate(match(v, uniqv)))]
}

# 
populate_ind_barcodes <- function(i, ind_barcodes) {
  colors <- c("red", "green", "blue", "grey")
  for (j in 1:length(colors)) {
    color <- colors[j]
    vals <- data[data$barcode == row.names(ind_barcodes)[i],][,color]
    ind_barcodes[i, paste("mean_", color, sep="")] <- mean(vals)
    ind_barcodes[i, paste("median_", color, sep="")] <- median(vals)
    ind_barcodes[i, paste("mode_", color, sep="")] <- getmode(vals) 
  }
  ind_barcodes
}

# create empty df
ind_barcodes <- data.frame(matrix(ncol = 12, nrow = length(unique(data[,'barcode']))))
colnames(ind_barcodes) <- c('mean_red', 'median_red', 'mode_red', 'mean_green', 'median_green', 'mode_green', 'mean_blue', 'median_blue', 'mode_blue', 'mean_grey', 'median_grey', 'mode_grey')
rownames(ind_barcodes) <- unique(data[, 'barcode'])

# add mean, median, mode data
for (i in 1:nrow(ind_barcodes)) {
  ind_barcodes <- populate_ind_barcodes(i, ind_barcodes)
}
ind_barcodes

Plot Histogram of Color Channel Values at a given barcode

# histogram of color channel values in one spot
barcode = rownames(ind_barcodes)[1]
x <- data[data$barcode == barcode,'green']
hist(x, main="Histogram of Green Color Channel Values For a Single Barcode",breaks=100)

Function to Compute Corr Matrix in Parallel

compute_corr_matrix <- function(expression_arr, color_arr) {
  foreach(i = 1:ncol(expression_arr),
  .combine = rbind,
  .packages = c('data.table', 'doParallel')) %dopar% {
    colName <- colnames(expression_arr)[i]
    df <- data.frame(round(cor(expression_arr[,i], color_arr, method = 'pearson', use="complete.obs"), 3))
    rownames(df) <- colName
    df
  }
}

Analysis 1: Compute Correlation First, Compute Outliers After

In the following blocks, the correlation of between all genes + pixel color channel values is calculated. Then, for the top performing genes, a plotting function is called in which spots corresponding to outlier gene expression values + spots corresponding to outlier pixel color channel values are removed.

# uncomment this line if each spot has one pixel
color_arr <- data[,6:ncol(data)]

# uncomment this line if each spot has multiple pixels
# color_arr <- ind_barcodes
correlation_matrix <- compute_corr_matrix(logcounts_151507_filtered, color_arr)

# set threshold and filter
threshold <- 0.3
correlation_matrix_filter <- correlation_matrix
filter <- as.data.frame(apply(abs(correlation_matrix_filter) >= threshold, 1, any))
correlation_matrix_filter <- correlation_matrix_filter[filter[,1],]

# rename rownames based on gene name instead of gene ID
for (i in 1:nrow(correlation_matrix_filter)) {
  g_name <- rownames(correlation_matrix_filter)[i]
  rownames(correlation_matrix_filter)[i] <- rowData(spe[g_name])$gene_name
}
correlation_matrix_filter

# plot  correlation matrix
ggcorrplot(correlation_matrix_filter, sig.level=0.01, lab_size = 4.5, p.mat = NULL,
           insig = c("pch", "blank"), pch = 1, pch.col = "black", pch.cex =1,
           tl.cex = 7) +
  theme(axis.text.x = element_text(margin=margin(-2,0,0,0)),
        axis.text.y = element_text(margin=margin(0,-2,0,0)),
        panel.grid.minor = element_line(size=7)) + 
  geom_tile(fill="white") +
  geom_tile(height=1, width=1)

# function to calculate regression equation. Used as a sanity check
lm_eqn <- function(df){
    m <- lm(mean_red ~ MTRNR2L8, df);
    eq <- substitute(italic(y) == a + b %.% italic(x)*","~~italic(r)^2~"="~r2, 
         list(a = format(unname(coef(m)[1]), digits = 2),
              b = format(unname(coef(m)[2]), digits = 2),
             r2 = format(summary(m)$r.squared, digits = 3)))
    as.character(as.expression(eq));
}

# plot color channel value vs gene expression. If remove_outliers == TRUE, outlier gene expression values + outlier pixel color values are removed.
# ASSUMES THAT EACH BARCODE HAS MULTIPLE PIXELS
plot_channel_highres <- function(g_name, gene_id, color, stat, log_arr, color_arr, remove_outliers) {
  stat_type <- paste(color, stat, sep="_")
  x_axis <- log_arr[,gene_id]
  y_axis <- color_arr[,stat_type] * 255
  df <- data.frame(x_axis, y_axis)
  if (remove_outliers) {
    barcode_names <- rownames(log_arr)
    overlap <- union(rownames(log_arr)[isOutlier(log_arr[,gene_id])], rownames(color_arr)[isOutlier(color_arr[,stat_type])])
    rownames(df) <- barcode_names
    df <- df[-which(rownames(df) %in% overlap),]
    title <- paste(stat_type, " Channel Values vs ", g_name, " Gene Expression with Outliers Removed", sep="")
  } else {
    title <- paste(stat_type, " Channel Values vs ", g_name, " Gene Expression without Outliers Removed", sep="")
  }
  ggplot(df, aes(x=df[,1], y=df[,2])) + xlab(g_name) + ylab(stat_type) + geom_point(color=color) +
  geom_smooth(method='lm', color='black') + stat_regline_equation(label.y = 190, aes(label = ..eq.label..)) +
  stat_regline_equation(label.y = 180, aes(label = ..rr.label..)) + ggtitle(title) 
}

# call plot_channel function for all color and mean/median/mode combinations
plot_all_channels <- function(g_name, log_arr, color_arr, remove_outliers) {
  gene_id <- rowData(spe)[rowData(spe)$gene_name == g_name,]$gene_id
  colors <- c('red', 'green', 'blue', 'grey')
  stats <- c('mean', 'median', 'mode')

  for (color in colors) {
    for (stat in stats) {
      print(plot_channel_highres(g_name, gene_id, color, stat, log_arr, color_arr, remove_outliers)) 
    }
  }
}

# plot color channel value vs gene expression. If remove_outliers == TRUE, outlier gene expression values + outlier pixel color values are removed.
# ASSUMES THAT EACH BARCODE HAS ONE PIXEL
plot_channel <- function(g_name, gene_id, color, log_arr, color_arr, remove_outliers) {
  x_axis <- log_arr[,gene_id]
  y_axis <- color_arr[,color] * 255
  df <- data.frame(x_axis, y_axis)
  if (remove_outliers) {
    barcode_names <- rownames(log_arr)
    overlap <- union(rownames(log_arr)[isOutlier(log_arr[,gene_id])], rownames(color_arr)[isOutlier(color_arr[,color])])
    rownames(df) <- barcode_names
    df <- df[-which(rownames(df) %in% overlap),]
    title <- paste(color, " Channel Values vs ", g_name, " Gene Expression with Outliers Removed", sep="")
  } else {
    title <- paste(color, " Channel Values vs ", g_name, " Gene Expression without Outliers Removed", sep="")
  }
  ggplot(df, aes(x=df[,1], y=df[,2])) + xlab(g_name) + ylab(color) + geom_point(color=color) +
  geom_smooth(method='lm', color='black') + stat_regline_equation(label.y = 190, aes(label = ..eq.label..)) +
  stat_regline_equation(label.y = 180, aes(label = ..rr.label..)) + ggtitle(title) 
}

# call plot_channel function for all colors
plot_all_channels <- function(g_name, log_arr, color_arr, remove_outliers) {
  gene_id <- rowData(spe)[rowData(spe)$gene_name == g_name,]$gene_id
  colors <- c('red', 'green', 'blue', 'grey')

  for (color in colors) {
    print(plot_channel(g_name, gene_id, color, log_arr, color_arr, remove_outliers))
  }
}

plot_all_channels('MT-ND1', logcounts_151507_filtered, color_arr, remove_outliers=FALSE)

`geom_smooth()` using formula 'y ~ x'
`geom_smooth()` using formula 'y ~ x'
`geom_smooth()` using formula 'y ~ x'
`geom_smooth()` using formula 'y ~ x'

plot_all_channels('MT-ND1', logcounts_151507_filtered, color_arr, remove_outliers=TRUE)

`geom_smooth()` using formula 'y ~ x'
`geom_smooth()` using formula 'y ~ x'
`geom_smooth()` using formula 'y ~ x'
`geom_smooth()` using formula 'y ~ x'

plot_all_channels('MT-CO2', logcounts_151507_filtered, color_arr, remove_outliers=FALSE)

`geom_smooth()` using formula 'y ~ x'
`geom_smooth()` using formula 'y ~ x'
`geom_smooth()` using formula 'y ~ x'
`geom_smooth()` using formula 'y ~ x'

plot_all_channels('MT-CO2', logcounts_151507_filtered, color_arr, remove_outliers=TRUE)

`geom_smooth()` using formula 'y ~ x'
`geom_smooth()` using formula 'y ~ x'
`geom_smooth()` using formula 'y ~ x'
`geom_smooth()` using formula 'y ~ x'

plot_all_channels('MT-ATP6', logcounts_151507_filtered, color_arr, remove_outliers=FALSE)

`geom_smooth()` using formula 'y ~ x'
`geom_smooth()` using formula 'y ~ x'
`geom_smooth()` using formula 'y ~ x'
`geom_smooth()` using formula 'y ~ x'

plot_all_channels('MT-ATP6', logcounts_151507_filtered, color_arr, remove_outliers=TRUE)

`geom_smooth()` using formula 'y ~ x'
`geom_smooth()` using formula 'y ~ x'
`geom_smooth()` using formula 'y ~ x'
`geom_smooth()` using formula 'y ~ x'

plot_all_channels('MT-CYB', logcounts_151507_filtered, color_arr, remove_outliers=FALSE)

`geom_smooth()` using formula 'y ~ x'
`geom_smooth()` using formula 'y ~ x'
`geom_smooth()` using formula 'y ~ x'
`geom_smooth()` using formula 'y ~ x'

plot_all_channels('MT-CYB', logcounts_151507_filtered, color_arr, remove_outliers=TRUE)

`geom_smooth()` using formula 'y ~ x'
`geom_smooth()` using formula 'y ~ x'
`geom_smooth()` using formula 'y ~ x'
`geom_smooth()` using formula 'y ~ x'

Redo Analysis Removing Outliers Before Correlation Computation

# Remove outlier gene expression values. If median gene expression value is 0, replace complement of outliers with NA. Else replace outliers with NA. If remaining non-NA expression values < num, replace entire column with NA.
remove_outliers <- function(arr, num) {
  foreach(i = 1:ncol(arr),
  .combine = cbind,
  .packages = c('data.table', 'doParallel')) %dopar% {
    if (median(arr[,i]) == 0) {
      arr[,i][!isOutlier(arr[,i])] <- NA
    } else {
      arr[,i][isOutlier(arr[,i])] <- NA
    }
    if (matrixStats::count(!is.na(arr[,i])) < num) {
      arr[,i] <- rep(NA, nrow(arr))
    }
    arr[,i]
  }
}

# Call remove outlier function
logcounts_temp <- as.matrix(logcounts_151507_filtered)
col_names <- colnames(logcounts_temp)
logcounts_151507_filtered_no_outliers <- remove_outliers(logcounts_temp, 100)
colnames(logcounts_151507_filtered_no_outliers) <- col_names
logcounts_151507_filtered_no_outliers

# Remove columns that are completely NA
not_all_na <- function(x) any(!is.na(x))
logcounts_151507_filtered_no_outliers <- data.frame(logcounts_151507_filtered_no_outliers) %>% select(where(not_all_na))

dim(logcounts_151507_filtered_no_outliers)

[1] 4165 9807

# compute correlation matrix
mat <- compute_corr_matrix(logcounts_151507_filtered_no_outliers, color_arr)

# filter corr matrix based on threshold
threshold <- 0.35
correlation_matrix_filter <- mat
filter <- as.data.frame(apply(abs(correlation_matrix_filter) >= threshold, 1, any))
correlation_matrix_filter <- na.omit(correlation_matrix_filter[filter[,1],])
correlation_matrix_filter

# rename rows of correlation matrix with gene name
for (i in 1:nrow(correlation_matrix_filter)) {
  g_name <- rownames(correlation_matrix_filter)[i]
  print(g_name)
  rownames(correlation_matrix_filter)[i] <- rowData(spe[g_name])$gene_name
}

[1] "ENSG00000134201"
[1] "ENSG00000158711"
[1] "ENSG00000186205"
[1] "ENSG00000205795"
[1] "ENSG00000091409"
[1] "ENSG00000187288"
[1] "ENSG00000114698"
[1] "ENSG00000163132"
[1] "ENSG00000164066"
[1] "ENSG00000106003"
[1] "ENSG00000148204"
[1] "ENSG00000134817"
[1] "ENSG00000177283"
[1] "ENSG00000025423"
[1] "ENSG00000051825"
[1] "ENSG00000140022"
[1] "ENSG00000217930"
[1] "ENSG00000141569"
[1] "ENSG00000167676"
[1] "ENSG00000183579"
[1] "ENSG00000198712"

correlation_matrix_filter

# Correlation plot visualization
ggcorrplot(correlation_matrix_filter, sig.level=0.01, lab_size = 4.5, p.mat = NULL,
           insig = c("pch", "blank"), pch = 1, pch.col = "black", pch.cex =1,
           tl.cex = 7) +
  theme(axis.text.x = element_text(margin=margin(-2,0,0,0)),
        axis.text.y = element_text(margin=margin(0,-2,0,0)),
        panel.grid.minor = element_line(size=7)) + 
  geom_tile(fill="white") +
  geom_tile(height=1, width=1)

plot_all_channels('MARC1', logcounts_151507_filtered_no_outliers, color_arr, remove_outliers=FALSE)

`geom_smooth()` using formula 'y ~ x'
Warning: Removed 4049 rows containing non-finite values (stat_smooth).
Warning: Removed 4049 rows containing non-finite values (stat_regline_equation).
Warning: Removed 4049 rows containing non-finite values (stat_regline_equation).
Warning: Removed 4049 rows containing missing values (geom_point).
`geom_smooth()` using formula 'y ~ x'
Warning: Removed 4049 rows containing non-finite values (stat_smooth).
Warning: Removed 4049 rows containing non-finite values (stat_regline_equation).
Warning: Removed 4049 rows containing non-finite values (stat_regline_equation).
Warning: Removed 4049 rows containing missing values (geom_point).
`geom_smooth()` using formula 'y ~ x'
Warning: Removed 4049 rows containing non-finite values (stat_smooth).
Warning: Removed 4049 rows containing non-finite values (stat_regline_equation).
Warning: Removed 4049 rows containing non-finite values (stat_regline_equation).
Warning: Removed 4049 rows containing missing values (geom_point).
`geom_smooth()` using formula 'y ~ x'
Warning: Removed 4049 rows containing non-finite values (stat_smooth).
Warning: Removed 4049 rows containing non-finite values (stat_regline_equation).
Warning: Removed 4049 rows containing non-finite values (stat_regline_equation).
Warning: Removed 4049 rows containing missing values (geom_point).

plot_all_channels('HSD17B6', logcounts_151507_filtered_no_outliers, color_arr, remove_outliers=FALSE)

`geom_smooth()` using formula 'y ~ x'
Warning: Removed 4042 rows containing non-finite values (stat_smooth).
Warning: Removed 4042 rows containing non-finite values (stat_regline_equation).
Warning: Removed 4042 rows containing non-finite values (stat_regline_equation).
Warning: Removed 4042 rows containing missing values (geom_point).
`geom_smooth()` using formula 'y ~ x'
Warning: Removed 4042 rows containing non-finite values (stat_smooth).
Warning: Removed 4042 rows containing non-finite values (stat_regline_equation).
Warning: Removed 4042 rows containing non-finite values (stat_regline_equation).
Warning: Removed 4042 rows containing missing values (geom_point).
`geom_smooth()` using formula 'y ~ x'
Warning: Removed 4042 rows containing non-finite values (stat_smooth).
Warning: Removed 4042 rows containing non-finite values (stat_regline_equation).
Warning: Removed 4042 rows containing non-finite values (stat_regline_equation).
Warning: Removed 4042 rows containing missing values (geom_point).
`geom_smooth()` using formula 'y ~ x'
Warning: Removed 4042 rows containing non-finite values (stat_smooth).
Warning: Removed 4042 rows containing non-finite values (stat_regline_equation).
Warning: Removed 4042 rows containing non-finite values (stat_regline_equation).
Warning: Removed 4042 rows containing missing values (geom_point).

plot_all_channels('MSX1', logcounts_151507_filtered_no_outliers, color_arr, remove_outliers=FALSE)

`geom_smooth()` using formula 'y ~ x'
Warning: Removed 3996 rows containing non-finite values (stat_smooth).
Warning: Removed 3996 rows containing non-finite values (stat_regline_equation).
Warning: Removed 3996 rows containing non-finite values (stat_regline_equation).
Warning: Removed 3996 rows containing missing values (geom_point).
`geom_smooth()` using formula 'y ~ x'
Warning: Removed 3996 rows containing non-finite values (stat_smooth).
Warning: Removed 3996 rows containing non-finite values (stat_regline_equation).
Warning: Removed 3996 rows containing non-finite values (stat_regline_equation).
Warning: Removed 3996 rows containing missing values (geom_point).
`geom_smooth()` using formula 'y ~ x'
Warning: Removed 3996 rows containing non-finite values (stat_smooth).
Warning: Removed 3996 rows containing non-finite values (stat_regline_equation).
Warning: Removed 3996 rows containing non-finite values (stat_regline_equation).
Warning: Removed 3996 rows containing missing values (geom_point).
`geom_smooth()` using formula 'y ~ x'
Warning: Removed 3996 rows containing non-finite values (stat_smooth).
Warning: Removed 3996 rows containing non-finite values (stat_regline_equation).
Warning: Removed 3996 rows containing non-finite values (stat_regline_equation).
Warning: Removed 3996 rows containing missing values (geom_point).

plot_all_channels('MT-CO2', logcounts_151507_filtered_no_outliers, color_arr, remove_outliers=FALSE)

`geom_smooth()` using formula 'y ~ x'
Warning: Removed 20 rows containing non-finite values (stat_smooth).
Warning: Removed 20 rows containing non-finite values (stat_regline_equation).
Warning: Removed 20 rows containing non-finite values (stat_regline_equation).
Warning: Removed 20 rows containing missing values (geom_point).
`geom_smooth()` using formula 'y ~ x'
Warning: Removed 20 rows containing non-finite values (stat_smooth).
Warning: Removed 20 rows containing non-finite values (stat_regline_equation).
Warning: Removed 20 rows containing non-finite values (stat_regline_equation).
Warning: Removed 20 rows containing missing values (geom_point).
`geom_smooth()` using formula 'y ~ x'
Warning: Removed 20 rows containing non-finite values (stat_smooth).
Warning: Removed 20 rows containing non-finite values (stat_regline_equation).
Warning: Removed 20 rows containing non-finite values (stat_regline_equation).
Warning: Removed 20 rows containing missing values (geom_point).
`geom_smooth()` using formula 'y ~ x'
Warning: Removed 20 rows containing non-finite values (stat_smooth).
Warning: Removed 20 rows containing non-finite values (stat_regline_equation).
Warning: Removed 20 rows containing non-finite values (stat_regline_equation).
Warning: Removed 20 rows containing missing values (geom_point).

plot_all_channels('PLIN4', logcounts_151507_filtered_no_outliers, color_arr, remove_outliers=FALSE)

`geom_smooth()` using formula 'y ~ x'
Warning: Removed 3969 rows containing non-finite values (stat_smooth).
Warning: Removed 3969 rows containing non-finite values (stat_regline_equation).
Warning: Removed 3969 rows containing non-finite values (stat_regline_equation).
Warning: Removed 3969 rows containing missing values (geom_point).
`geom_smooth()` using formula 'y ~ x'
Warning: Removed 3969 rows containing non-finite values (stat_smooth).
Warning: Removed 3969 rows containing non-finite values (stat_regline_equation).
Warning: Removed 3969 rows containing non-finite values (stat_regline_equation).
Warning: Removed 3969 rows containing missing values (geom_point).
`geom_smooth()` using formula 'y ~ x'
Warning: Removed 3969 rows containing non-finite values (stat_smooth).
Warning: Removed 3969 rows containing non-finite values (stat_regline_equation).
Warning: Removed 3969 rows containing non-finite values (stat_regline_equation).
Warning: Removed 3969 rows containing missing values (geom_point).
`geom_smooth()` using formula 'y ~ x'
Warning: Removed 3969 rows containing non-finite values (stat_smooth).
Warning: Removed 3969 rows containing non-finite values (stat_regline_equation).
Warning: Removed 3969 rows containing non-finite values (stat_regline_equation).
Warning: Removed 3969 rows containing missing values (geom_point).

plot_all_channels('CIDEC', logcounts_151507_filtered_no_outliers, color_arr, remove_outliers=FALSE)

`geom_smooth()` using formula 'y ~ x'
Warning: Removed 4063 rows containing non-finite values (stat_smooth).
Warning: Removed 4063 rows containing non-finite values (stat_regline_equation).
Warning: Removed 4063 rows containing non-finite values (stat_regline_equation).
Warning: Removed 4063 rows containing missing values (geom_point).
`geom_smooth()` using formula 'y ~ x'
Warning: Removed 4063 rows containing non-finite values (stat_smooth).
Warning: Removed 4063 rows containing non-finite values (stat_regline_equation).
Warning: Removed 4063 rows containing non-finite values (stat_regline_equation).
Warning: Removed 4063 rows containing missing values (geom_point).
`geom_smooth()` using formula 'y ~ x'
Warning: Removed 4063 rows containing non-finite values (stat_smooth).
Warning: Removed 4063 rows containing non-finite values (stat_regline_equation).
Warning: Removed 4063 rows containing non-finite values (stat_regline_equation).
Warning: Removed 4063 rows containing missing values (geom_point).
`geom_smooth()` using formula 'y ~ x'
Warning: Removed 4063 rows containing non-finite values (stat_smooth).
Warning: Removed 4063 rows containing non-finite values (stat_regline_equation).
Warning: Removed 4063 rows containing non-finite values (stat_regline_equation).
Warning: Removed 4063 rows containing missing values (geom_point).

plot_all_channels('ELK4', logcounts_151507_filtered_no_outliers, color_arr, remove_outliers=FALSE)

`geom_smooth()` using formula 'y ~ x'
Warning: Removed 4043 rows containing non-finite values (stat_smooth).
Warning: Removed 4043 rows containing non-finite values (stat_regline_equation).
Warning: Removed 4043 rows containing non-finite values (stat_regline_equation).
Warning: Removed 4043 rows containing missing values (geom_point).
`geom_smooth()` using formula 'y ~ x'
Warning: Removed 4043 rows containing non-finite values (stat_smooth).
Warning: Removed 4043 rows containing non-finite values (stat_regline_equation).
Warning: Removed 4043 rows containing non-finite values (stat_regline_equation).
Warning: Removed 4043 rows containing missing values (geom_point).
`geom_smooth()` using formula 'y ~ x'
Warning: Removed 4043 rows containing non-finite values (stat_smooth).
Warning: Removed 4043 rows containing non-finite values (stat_regline_equation).
Warning: Removed 4043 rows containing non-finite values (stat_regline_equation).
Warning: Removed 4043 rows containing missing values (geom_point).
`geom_smooth()` using formula 'y ~ x'
Warning: Removed 4043 rows containing non-finite values (stat_smooth).
Warning: Removed 4043 rows containing non-finite values (stat_regline_equation).
Warning: Removed 4043 rows containing non-finite values (stat_regline_equation).
Warning: Removed 4043 rows containing missing values (geom_point).

plot_all_channels('LFNG', logcounts_151507_filtered_no_outliers, color_arr, remove_outliers=FALSE)

`geom_smooth()` using formula 'y ~ x'
Warning: Removed 4059 rows containing non-finite values (stat_smooth).
Warning: Removed 4059 rows containing non-finite values (stat_regline_equation).
Warning: Removed 4059 rows containing non-finite values (stat_regline_equation).
Warning: Removed 4059 rows containing missing values (geom_point).
`geom_smooth()` using formula 'y ~ x'
Warning: Removed 4059 rows containing non-finite values (stat_smooth).
Warning: Removed 4059 rows containing non-finite values (stat_regline_equation).
Warning: Removed 4059 rows containing non-finite values (stat_regline_equation).
Warning: Removed 4059 rows containing missing values (geom_point).
`geom_smooth()` using formula 'y ~ x'
Warning: Removed 4059 rows containing non-finite values (stat_smooth).
Warning: Removed 4059 rows containing non-finite values (stat_regline_equation).
Warning: Removed 4059 rows containing non-finite values (stat_regline_equation).
Warning: Removed 4059 rows containing missing values (geom_point).
`geom_smooth()` using formula 'y ~ x'
Warning: Removed 4059 rows containing non-finite values (stat_smooth).
Warning: Removed 4059 rows containing non-finite values (stat_regline_equation).
Warning: Removed 4059 rows containing non-finite values (stat_regline_equation).
Warning: Removed 4059 rows containing missing values (geom_point).

dim(logcounts_temp)

[1]  4165 21134

dim(logcounts_151507_filtered_no_outliers)

[1] 4165 9807

Find Location of Outliers on Histology Image

# plot_outliers_in_lowres <- function(g_name, stat_type) {
#   overlap <- union(rownames(max_ex_corr)[isOutlier(max_ex_corr[,g_name])], rownames(ind_barcodes)[isOutlier(ind_barcodes[,stat_type])])
#   x_axis <- data[which(data$barcode %in% overlap),]$scaled_pxl_col_in_lowres
#   y_axis <- data[which(data$barcode %in% overlap),]$scaled_pxl_row_in_lowres
#   
#   ggplot(data.frame(x_axis, y_axis), aes(x=x_axis, y=y_axis)) + xlab('Scaled Pixel Column in LowRes') + ylab('Scaled Pixel Row in LowRes') + geom_point(color='navy') +
#     ggtitle(paste('Location of',stat_type, g_name, "Outliers on LowRes Image", sep=" ")) 
# }
# ```
# 
# 
# ```{r}
# plot_outliers_in_lowres('MT3', 'mean_red')
# plot_outliers_in_lowres('MT3', 'mean_blue')
# plot_outliers_in_lowres('MT3', 'mean_green')
# ```
# 
# ```{r}
# plot_outliers_in_lowres('MTRNR2L8', 'mean_red')
# plot_outliers_in_lowres('MTRNR2L8', 'mean_blue')
# plot_outliers_in_lowres('MTRNR2L8', 'mean_green')
# ```